Download - Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

Transcript
Page 1: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

Do startups provide employment opportunities for disadvantaged workers?

Daniel FacklerInstitut für Wirtschaftsforschung Halle

Michaela FuchsIAB Sachsen-Anhalt-Thüringen

Lisa HölscherInstitut für Wirtschaftsforschung Halle

Claus SchnabelUniversity of Erlangen-Nuremberg

(June 2018)

LASER Discussion Papers - Paper No. 108

(edited by A. Abele-Brehm, R.T. Riphahn, K. Moser and C. Schnabel)

Correspondence to:

Claus Schnabel, Lange Gasse 20, 90020 Nuremberg, Germany, Email:[email protected].

Page 2: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

Abstract

This paper analyzes whether startups offer job opportunities to workers potentially facing labor marketproblems. It compares the hiring patterns of startups and incumbents in the period 2003 to 2014 usingadministrative linked employer-employee data for Germany that allow to take the completeemployment biographies of newly hired workers into account. The results indicate that young plantsare more likely than incumbents to hire older and foreign applicants as well as workers who haveinstable employment biographies, come from unemployment or outside the labor force, or wereaffected by a plant closure. However, an analysis of entry wages reveals that disadvantageous workercharacteristics come along with higher wage penalties in startups than in incumbents. Therefore, evenif startups provide employment opportunities for certain groups of disadvantaged workers, the qualityof these jobs in terms of initial remuneration seems to be low.

Zusammenfassung

Die Studie analysiert, ob neu gegründete Betriebe Beschäftigungsmöglichkeiten für solcheArbeitnehmer bieten, die zu den Problemgruppen des Arbeitsmarktes zählen. Unter Verwendungadministrativer, verbundener Arbeitgeber-Arbeitnehmer-Daten für Deutschland, die eineBerücksichtigung der gesamten Erwerbsbiografien von neu eingestellten Arbeitnehmernermöglichen, vergleicht sie die Einstellungsmuster von neu gegründeten und etablierten Betriebenim Zeitraum 2003-2014. Es zeigt sich, dass junge Betriebe tatsächlich mit einer höherenWahrscheinlichkeit als etablierte Betriebe ältere und ausländische Arbeitnehmer sowie solche mitinstabilen Erwerbsbiografien einstellen. Gleiches gilt für Bewerber, die aus Arbeitslosigkeit odervon außerhalb des Arbeitsmarktes kommen oder die Opfer einer Betriebsschließung wurden.Allerdings deutet eine Analyse der Einstiegslöhne darauf hin, dass die Merkmale dieserbenachteiligten Arbeitnehmer in neu gegründeten Betrieben mit höheren Lohnabschlägeneinhergehen als in etablierten Betrieben. Auch wenn Neugründungen damitBeschäftigungsmöglichkeiten für bestimmte Gruppen benachteiligter Arbeitnehmer bieten,scheint die Qualität dieser Jobs � gemessen an der anfänglichen Entlohnung � gering zu sein.

Author note

We would like to thank Udo Brixy and Steffen Müller, seminar participants at the Halle Institute forEconomic Research (IWH), the Otto von Guericke University Magdeburg, and the Colloquium onPersonnel Economics (COPE) 2018 in Munich for helpful comments and suggestions.

Page 3: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

1. INTRODUCTION

In political debates, startups are often regarded as important drivers of structural change and

technological progress and they are ascribed a crucial role for job creation, thereby helping

to reduce unemployment. It is thus not surprising that a broad literature has dealt with newly

founded firms, their performance and their contribution to job creation and destruction.1

What is surprising, however, is that there is not much empirical evidence on the actual hiring

behavior of newly founded firms. This research deficit is particularly grave because the

relevance of startups and their direct contribution to overcoming employment problems will

be larger if they disproportionally hire workers who are currently not employed, who have

difficulties finding jobs in mature firms or who lost their jobs in the course of reallocation

and structural change (e.g., due to plant closures). Even if the jobs in startups are less stable

than those in incumbent firms, they may still help to preserve workers’ labor market

attachment, prevent human capital depreciations coming along with longer periods of

unemployment, and make it easier for work seekers to re-enter the labor market. In contrast,

if startups just poach workers from incumbent firms, they mainly contribute to labor market

turnover. In this case, it is questionable whether their direct contribution to overcoming

employment problems of certain groups of workers is substantial enough to warrant the

strong political attention and support startups currently receive.

Against this background, the primary objective of this paper is to analyze empirically

whether startups are more likely than incumbent firms to provide employment opportunities

for so-called “disadvantaged” workers facing serious labor market problems – in particular

older workers, foreigners, low-qualified individuals, persons with unstable employment

biographies, in (long-term) unemployment or outside the labor force, as well as first-time

entrants into the labor market and workers who have become victims of plant closures.

Startups may offer such workers a riskier and probably lower-paying alternative when being

shut out of jobs at mature firms (an alternative that is still better than being unemployed),

but in their critical early phase, these newly founded firms could also be reluctant to recruit

individuals with obvious deficiencies. If startups are found to be more likely to provide

employment opportunities for disadvantaged workers, this implies that they are not only

beneficial for an economy by fostering growth and competition, but also that the jobs created

by them are valuable from a socio-political point of view.2 We add to the literature not only

1 Surveys of the literature on newly founded firms are provided by Geroski (1995), Wagner (2006)

or Santarelli and Vivarelli (2007).

2 Our analysis focuses on whether startups themselves directly contribute to overcoming

employment problems by hiring disadvantaged workers. We are aware that even if startups are

poaching workers with more desirable characteristics from established firms, this might lead to a

redeployment process in which these vacant positions in incumbents could be filled with

Page 4: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

2

by focusing on workers with labor market problems but also by making use of more detailed

information about workers’ employment biographies than previous studies as we have access

to high-quality linked employer-employee data for (West) Germany reaching back to 1975.

Beyond the analysis of job opportunities for disadvantaged workers, we additionally address

the quality of these jobs by investigating whether the above-mentioned worker

characteristics come along with wage penalties and whether these penalties are higher in

young or incumbent firms. If they are lower in young firms, for instance because startups

are not willing or able to discriminate against certain types of workers or assess these

workers’ human capital differently than incumbent firms, startups provide an additional

pecuniary benefit for disadvantaged workers. In contrast, it could also be argued that it is

incumbents that have less scope for discrimination than startups due to wage setting

institutions like collective agreements and works councils. To the best of our knowledge, we

are the first to analyze wage differentials between startups and incumbents specifically for

disadvantaged groups of workers.

2. EMPLOYMENT, HIRING BEHAVIOR, AND WAGES IN STARTUPS

Many studies have shown that young firms’ contribution to gross and net job creation is

substantial (see, e.g., Haltiwanger, Jarmin, and Miranda 2013 for the US; Fuchs and Weyh

2010 for Germany). At the same time, young firms also contribute disproportionately to job

destruction, in particular because of their high exit rates (e.g., Fackler, Schnabel, and Wagner

2013). Using data for Germany, Fritsch and Weyh (2006) demonstrate that the total number

of jobs in a startup cohort first increases but then falls below its initial level after a couple of

years, mainly because many of these startups exit the market.3 Some authors therefore

question whether startups really play an important role for sustainable job creation (e.g.,

Santarelli and Vivarelli 2007; Shane 2009).

Despite this growing and controversial literature on the overall employment effects of

startups, there is not much empirical evidence on the actual hiring behavior of newly founded

firms, as observed by Fairlie and Miranda (2017, p. 3): “Job creation is one of the most

important aspects of entrepreneurship, but we know relatively little about the hiring patterns

disadvantaged individuals. Analyzing these dynamics in detail is however beyond the scope of

our study.

3 In addition to these direct employment effects, indirect effects might emerge from the increased

competitive pressure exerted by startups, which induces incumbents to react, thereby fostering

economic growth. Fritsch and Noseleit (2013), for example, find for Germany that this indirect

employment effect of startups is substantial, too.

Page 5: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

3

and decisions of startups.”4 Although there are some studies addressing various aspects of

young firms’ hiring behavior, the extant literature is quite small and the relevance of startups

for disadvantaged workers has not been its main research question.

Using data for Sweden, Nyström (2012) shows that immigrants and labor market entrants

are more likely and women are less likely to be hired by new firms. Also for Sweden,

Nyström and Elvung (2015) analyze wage penalties in startups for voluntary versus

involuntary job switchers. They report that employees who have to switch jobs because of

firm closures are more likely to end up in startups. As a byproduct of investigating

employment stability in newly founded firms, a study for Germany by Schnabel, Kohaut,

and Brixy. (2011) also provides some evidence on the characteristics of individuals joining

these firms. The authors find, inter alia, that individuals who had more jobs or a larger

number of unemployment spells are more likely to join newly founded firms, whereas the

opposite is true for workers with longer employment experience. In a study for Denmark,

Coad, Nielsen, and Timmermans (2017) look at the effects of solo entrepreneurs’ decision

to hire their very first employee on their sales and profits, but they also report that “more

marginalized” workers (such as older or previously unemployed individuals) have a higher

probability of becoming a new firm’s first employee. Finally, focusing on workers’ age,

Ouimet and Zarutskie (2014) show for the US that young firms disproportionally hire young

workers.5

Our paper contributes to this small literature, replicating some of the prior results and

questioning others, but it goes beyond previous studies, which have reported results for

selected worker characteristics in isolation. We will put special emphasis on various groups

of workers facing labor market problems, employing a battery of disadvantaged workers’

characteristics. In doing so, we are able to use a more comprehensive data set and more

detailed information about workers’ employment biographies than previous studies as we

have access to high-quality linked employer-employee data reaching back to 1975. In

addition, we address the quality of these jobs by investigating whether these adverse worker

characteristics come along with wage penalties and whether these penalties are higher in

young or incumbent firms.

4 Fairlie and Miranda (2017) study under which circumstances newly founded firms start hiring

employees, but they do not address the question which types of workers are hired.

5 Beyond that, some studies have investigated the relationship between the composition of the

initial workforce (e.g., in terms of gender, age or qualification) and firm performance in terms of

survival or growth (e.g., Weber and Zulehner 2010, 2014; Geroski, Mata, and Portugal 2010;

Koch, Späth, and Strotmann 2013). Other studies have analyzed the role of founders or the

importance of the initial human capital of founders and employees for the success of startups

(e.g., Brüderl, Preisendörfer, and Ziegler 2007; Dahl and Reichstein 2007; Rocha, van Praag,

Folta, and Carneiro 2016). However, these studies do not address the hiring patterns of young

firms in detail nor compare them to incumbent firms.

Page 6: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

4

Although some studies have already analyzed wage differentials between startups and

incumbents, extant studies do not provide a clear picture on whether workers (and in

particular disadvantaged workers) are better or worse off when joining startups compared to

incumbent firms. The majority of extant studies find that new firms tend to pay lower wages

(see, e.g., Brixy, Kohaut, and Schnabel 2007 for Germany; Nyström and Elvung 2014 for

Sweden). In a detailed analysis of Danish registry data, Burton, Dahl, and Sorenson (2017)

observe both firm age and firm size effects when controlling for employee characteristics.

They find that typically startups pay less than mature employers, but the largest startups even

pay a wage premium. In contrast, using US data, Brown and Medoff (2003) report no

significant wage differences, and Ouimet and Zarutskie (2014) even detect a wage premium

in startups for young workers and for new hires. Similarly, based on linked employer-

employee data for Germany, Schmieder (2013) finds that new establishments pay

significantly higher starting wages than establishments that are older than 20 years. Our

study contributes to this literature and will confirm that startups pay lower wages. It goes

beyond extant studies by analyzing wage differentials between startups and incumbents

specifically for disadvantaged groups of workers and by showing how workers’

characteristics and establishment age interact in wage determination.

3. THEORETICAL CONSIDERATIONS AND EXTANT EMPIRICAL EVIDENCE

In our analysis, we focus on the employment opportunities in startups compared to

incumbent firms with specific respect to several groups of disadvantaged workers who are

usually most affected by unemployment and who may have serious problems of (re-)entering

the labor market.6 In particular, we look at eight employment-inhibiting characteristics of

individuals and investigate whether workers with these characteristics are more or less likely

to be employed by startups. The first three characteristics are age above 50 years, foreign

nationality, and low qualification, since the respective groups of persons experience above-

average unemployment rates in Germany (Bundesagentur für Arbeit 2016). Related, we also

look at workers with unstable employment biographies who have received unemployment

benefits during a relatively high proportion of their working life, which may be a negative

signal to potential employers. Additionally, we take account of the origin or previous labor

market state of individuals hired, specifically focusing on whether they come from

unemployment or from outside the labor force, which may reduce their employment

prospects because of the loss of human capital associated with employment gaps. We further

consider first-time entrants to the labor market whose lack of work experience may make it

6 See, e.g., Möller and Walwei (2017) for a recent overview of the German labor market.

Page 7: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

5

more difficult to find a job. Finally, although they do not possess disadvantageous

characteristics per se, we also include workers who have become victims of plant closure of

their last employers since they are often found to experience severe and long-lasting

consequences of job loss (see, e.g., Fackler and Hank 2016).

Although there is no elaborate theory of individuals’ decision to join startups rather than

incumbent firms and of startups’ hiring decisions, we can build on some arguments and

insights from labor economics, industrial organization and entrepreneurship research to

derive testable hypotheses on the employment of disadvantaged workers in startups. Taking

first the perspective of the employer, startups are confronted with several fundamental

problems that make it difficult to attract employees. First, newly founded firms usually do

not have much experience in recruiting employees and may thus be at a disadvantage

compared to older and larger firms which have expert personnel departments and can also

rely on their name and reputation to attract talented workers (Nyström and Elvung 2015).

Second, startups and young firms have a higher risk of failure than incumbents (Fackler et

al. 2013), which implies that they should have to compensate workers for the higher risk of

job loss. This makes it costlier to attract employees, ceteris paribus. Third, startups typically

operate at such a small scale of output that they incur an inherent cost disadvantage and they

also face tighter financial constraints than older firms so that they must pursue a strategy of

compensating factor differentials, which includes paying lower wages (Audretsch, van

Leeuwen, Menkveld, and Thurik 2001; Michelacci and Quadrini 2005). For these reasons,

startups may find it difficult to poach employees from other firms but may have to rely more

on attracting individuals who are currently unemployed or outside the labor force (Schnabel

et al. 2011; Coad et al. 2017).

This reasoning also applies to individuals who enter the labor market for the first time, but

since labor market entrants do not possess working experience, newly founded firms that are

lacking established work routines and are more reliant on their employees’ expertise may

hesitate to hire them. In contrast, experienced workers can be recruited among workers who

recently lost their jobs in plant closures. Since it is difficult to attract first-class prime age

workers, newly founded firms might also have to recruit among “marginalized” workers

(Coad et al. 2017), i.e., groups with labor market problems such as older workers, individuals

with non-German nationality, low-qualified workers, and workers with instable employment

biographies. However, as the first hiring decision(s) can be crucial for the success and

survival of startups (Koch et al. 2013; Rocha et al. 2016), newly founded firms may be

reluctant to recruit individuals with obvious deficiencies such as low-qualified workers or

workers with perforated employment histories, at least in their critical early phase.

From the perspective of the employee, the decision to take up a job in a newly founded firm

(rather than joining an incumbent firm or being unemployed) is based on a comparison of

Page 8: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

6

the monetary and non-monetary returns with the risks and mobility costs from working there.

Employees will join only if their expected discounted lifetime utility is higher in startups,

which probably will not be the case for many workers. For instance, labor market entrants

coming from the educational system may hesitate to join startups that exhibit a high risk of

failure because the first job can be an important determinant of future success in the labor

market (Schnabel et al. 2011). A similar reasoning may apply to individuals who are

currently out of the labor force and re-enter the labor market.

On the other hand, employees who are unemployed, outside the labor force or who have had

instable employment biographies may risk working in a startup, even if this means lower

wages and higher employment instability compared to mature firms. One important reason

could be that in Germany any job that lasts at least 12 months entitles individuals to draw

unemployment benefits (again). Similarly, older (unemployed) employees who only need a

bridge into the pension system may be satisfied with a job in a startup even if it cannot be

expected to last particularly long. It could also be argued that employees who lost their jobs

(e.g. due to plant closures) may have less favorable unobservable characteristics and thus

sort themselves into smaller or more unstable firms (Nyström and Elvung 2015). Startups

could also be promising employers for foreign workers: if these workers are discriminated

against by incumbent firms (as shown by Kaas and Manger 2012), they may be better off

with startups that can probably afford less to discriminate, e.g., because of lower profits or a

lack of monopsony power.

Similar considerations may pertain to other groups of workers whose unfavorable

characteristics are associated with wage penalties. If these penalties are lower in newly

founded firms, for instance because startups are not willing or able to discriminate against

certain types of workers or differently assess the human capital of these workers than

incumbent firms do, then startups are relatively more attractive employers for disadvantaged

workers. However, it could also be argued that workers with disadvantageous characteristics

are better protected against wage discrimination when choosing to work in incumbent firms.

These firms are more likely to have professionalized personnel departments and more

elaborate personnel regulations, and they are more often covered by collective agreements

and works councils that make discrimination more difficult. At the same time, individuals

with problematic characteristics and unemployed workers may not really have a choice but

to join startups due to limited outside options (Coad et al. 2017), i.e. because they are not

offered any decent jobs by mature firms and because their unemployment benefits are about

to run out. In this case, startups could exploit the precarious situation of these workers by

offering them even worse working conditions.

Finally, individuals may be attracted to jobs in new firms if they have a preference for the

specific job attributes provided by the entrepreneurial work setting in startups such as higher

Page 9: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

7

work autonomy, flatter organizational hierarchy, and less bureaucracy than in established

firms (Roach and Sauermann 2015). Related, for all groups of (disadvantaged) workers

joining startups may be enticing if they speculate that they are now first in line and thus in a

good position for a career within the newly founded firm (if it does not fail).

These theoretical considerations imply the positive or negative relationships between our

eight main variables of interest and the probability of employment of disadvantaged workers

in startups shown in Table 1. Although the perspectives of newly founded firms and of

workers do not always coincide, in most cases relatively clear predictions concerning the

employment in newly founded firms are possible. When investigating these relationships in

reduced-form estimations, however, we should keep in mind that our empirical findings are

the result of an interaction of supply and demand and that we will not be able to clearly

distinguish between the decisions of individuals and of startup firms.

(Table 1 about here)

As mentioned above, there exists only a sparse empirical literature of not more than five

studies that provide multivariate analyses on which employees are working for startups.7

Mainly taking the employer’s perspective, Ouimet and Zarutskie (2014) show for the U.S.

that young firms disproportionately employ and hire young workers, arguing that this may

be due to the skills and risk tolerance of these workers. In contrast, using Danish data and

focusing on the first employee hired by solo entrepreneurs, Coad et al. (2017) find that the

probability of being recruited in a new firm increases with age (albeit at a decreasing rate).

They also show that workers coming from unemployment or outside the labor force are more

likely to be hired by a startup, while the opposite is found for persons who were enrolled in

education before recruitment. With Swedish data, Nyström (2012) finds that the likelihood

of being hired by a newly founded firm is lower for women but higher for immigrants and

for first-time entrants to the labor market, where the latter result stands in contrast to the

findings by Coad et al. (2017). Also for Sweden, Nyström and Elvung (2015) report that

employees who have to switch jobs due to firm closures are more likely to end up in startups.

Finally, a study for Germany by Schnabel et al. (2011) provides some evidence on the

characteristics of individuals joining startups. Although the impact of some socio-

demographic characteristics is statistically insignificant or differs between eastern and

western Germany, it becomes clear that individuals who had more jobs or more

unemployment spells are more likely to join newly founded firms, whereas the opposite is

true for workers with longer employment experience.

7 Theoretical analyses which types of employees (with different abilities and assets) may be found

in young firms are provided by Dahl and Klepper (2015) and Dinlersoz, Hyatt, and Janicki (2016).

Page 10: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

8

All in all, the empirical insights from these studies are neither clear-cut nor sufficient to

answer our main research question on the role of startups in providing employment

opportunities for disadvantaged workers. What is more, there is no empirical evidence at all

concerning possible differences in the wage penalties of disadvantaged workers between

startups and incumbents.

4. DATA AND DESCRIPTIVE STATISTICS

To analyze hiring patterns and wages in startups, we use extensive administrative data for

Germany based on social security notifications provided by the Institute for Employment

Research (IAB). We combine two sources, namely the Integrated Employment Biographies

(IEB) and the Establishment History Panel (BHP), to create a comprehensive linked

employer-employee data set that allows us to distinguish reliably between startups and

incumbents and to observe the complete labor market biographies of all workers entering

these establishments.

Detailed daily information on the labor market biographies of all workers in West Germany

subject to social security contributions from 1975 to 2014 is collected in the IEB. Since 1992,

the data set also includes information on East Germany and since 1999 it comprises

marginally employed individuals as well. The IEB contains detailed and very reliable micro-

level information on employment, job-search status, benefit receipt, and participation in

active labor market policy measures, along with individual characteristics like age, gender,

education, and nationality.8 It should be noted that the data only includes information on

hired employees who are subject to social security contributions. This implicates that the

founders of firms are not included, and we are therefore not able to analyze relationships

between a founder’s human capital and the quality of her initial workforce, as is done, e.g.,

by Rocha et al. (2016). This is, however, a minor shortcoming since our analyses focus on

hired employees rather than entrepreneurs.

Information on employers is provided in the BHP, a yearly panel that contains all

establishments with at least one employee subject to social security contributions. It includes

information on establishment size, industry, location, and workforce composition as of June

30th of a given year (for more information on the BHP, see Schmucker et al. 2016). As the

focus of our analysis lies on young establishments, it is crucial to identify startups as reliably

as possible. Since the occurrence of a new establishment identifier in the panel could be due

8 For more detailed information on the IEB, see Antoni, Ganzer, and vom Berge (2016) who

provide a description of the Sample of the Integrated Labour Market Biographies (SIAB), a 2

percent random sample from the IEB.

Page 11: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

9

to mere changes of the identification number, we make use of information on worker flows

(Hethey-Maier and Schmieder 2013). By observing the fraction of initial employees that

have previously worked together in another establishment, it is possible to distinguish

between true and spurious entries.9

However, since establishments are defined as local production units in the BHP and

information at the firm level is not included, we are not able to distinguish clearly between

the foundation of new, independent firms and the opening of new branches of multi-plant

firms. To assess the importance of this deficit, we had a look at the IAB Establishment Panel,

a yearly survey of around 16,000 establishments in Germany that contains information on

affiliations to multi-plant firms (see Ellguth, Kohaut, and Möller 2014 for details on this

representative data set). Our analysis revealed that in our period of observation around

85 percent of establishments in western Germany are independent legal units that do not

belong to a multi-plant firm. With respect to the identification of startups, we also reduce

the risk of observing new branches of multi-plant firms by excluding all those establishments

that report more than 20 employees in their first year of business, as recommended by Fritsch

and Brixy (2004). To evaluate this procedure, we link those establishments from the BHP

that we classify as startups and that meet the further sample restrictions described below with

the IAB Establishment Panel. This analysis reveals that 93 percent of the establishments that

we classify as startups can be categorized as new firms and only 7 percent as branch plant

foundations by existing firms, which we consider reasonably low.10

For our analysis, we draw a 10 percent random sample of those establishments that are newly

founded in the period 1999 to 201411 and define all those in their first five years of business

as young establishments. Therefore, our final observation period covers the years 2003 to

2014. New establishments are a more narrowly defined subgroup only including plants in

their first year. To construct a control group of incumbents, we draw a 5 percent sample of

all establishments existing during the same period and only keep those that are 5 years or

older. In both samples, we exclude establishments in agriculture, energy and mining, and in

the public and non-profit sector. For the remaining establishments, we link information on

all employees from the IEB (also referring to June 30th) covering their complete labor market

biographies. To ensure that workers’ biographies can be traced back over a long time horizon

9 In the following, we will only define establishments as newly founded if not more than 30 percent

of their initial workforce has worked together in the same establishment in the year before, or if

their initial workforce consists of no more than 3 persons. This definition is in accordance with

the categories “new (small)“ and “new (mid & big)” by Hethey-Maier and Schmieder (2013).

10 Although the IAB Establishment Panel provides some additional information at the firm level, it

would not make much sense to use it for the analysis of startups’ hiring patterns since the overall

number of startups in the IAB Establishment Panel is small and new establishments are typically

not included in the survey in their first year of existence.

11 Since 1999, the data also include marginally employed individuals. Due to the structural break,

we only use the data from 1999 onwards.

Page 12: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

10

and are not strongly left-censored, we focus on West German establishments and on

employees who are not older than 30 years when they are first observed in the IEB.12 We

only include individuals who have been newly hired by an establishment in the respective

year of observation, i.e., those who were not observed there in the previous year.

(Table 2 about here)

A descriptive overview over the establishments in our final sample is given in Table 2. On

average, young (and especially new) establishments are smaller than incumbents and operate

more often in the tertiary and less often in the secondary sector than established plants. With

respect to workforce composition, differences are not so pronounced.

To gain first insights into the characteristics of newly hired employees in young, new and

incumbent plants, Table 3 lists various individual-level variables. Workers entering startups

are more often older than 30 years than those hired by incumbents, indicating that young

firms may indeed opt for more experienced workers or that older people face problems

finding a job at an incumbent plant. We also see that the shares of foreigners, female hires

and of medium-qualified employees are higher in startups than in incumbents.13

(Table 3 about here)

Comparing previous labor market status before recruitment in the current establishment

reveals that workers entering startups are slightly more often coming from unemployment

or from outside the labor force, suggesting that startups might offer an opportunity to re-

enter employment for those workers with limited opportunities on the labor market.14 As

expected, only few first-time entrants find their way into young and especially new plants,

12 Since information on East Germany is just available from 1992 onwards, we would only be able

to observe the complete employment biographies of very young East German workers. To avoid

any bias from pooling this selective sample for East Germany with West German workers of all

ages, we exclude all East German establishments. By excluding persons older than 30 when first

observed in the IEB, we intend to limit our analyses to those workers whose first appearance in

the data coincides with their labor market entry. Relaxing this age restriction somewhat changes

the composition of the workforce in our sample but does not change our insights.

13 All employees who neither have Abitur, a German A-level equivalent, nor completed vocational

training, are defined as low-qualified. High-qualified workers are those holding a university

degree. Note that the education variable is imputed according to Fitzenberger, Osikominu, and

Völter (2005), see also Schmucker et al. (2016).

14 We define direct job-to-job transitions as recruitment if individuals have left their previous job not

more than 90 days before entering their current employer, i.e. workers with rather short periods

of frictional non-employment are not regarded as hires from unemployment or from outside the

labor force. Workers are categorized as being recruited from unemployment if they have left their

previous job more than 90 days ago and in the meantime were registered as job seeker, received

benefit payments, or participated in labor market programs. They are defined as coming from

outside the labor force if they were not observed at all in the data for more than 90 days. As the

IEB contains only workers subject to social security, “from outside the labor force” can also mean

that these workers were previously self-employed or – very unlikely – civil servants.

Page 13: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

11

which is also consistent with the fact that incumbents employ more apprentices. On average,

individuals entering startups spent a larger part of their working lives receiving

unemployment benefits and had a slightly larger number of previous employers, pointing

towards perforated and less stable employment biographies.

The last two variables in Table 3 refer only to individuals with previous working experience

(as observed in our data). On average, the share of workers coming from an establishment

that closed in the same year in which the worker left is twice as high among those entering

young rather than incumbent plants.15 Workers entering the subgroup of new plants have

been affected even more often by a closure of their previous establishment. Moreover,

individuals who are hired by a startup more often come from a young firm than those hired

by incumbents, pointing towards a certain labor market segmentation.

5. ECONOMETRIC ANALYSIS

Descriptive evidence suggests that on average unfavorable worker characteristics are more

often found among new hires in startups than among workers entering incumbents. In the

following, we will investigate in a multivariate framework whether startups are still more

likely to hire certain groups of disadvantaged workers when conditioning on a large set of

control variables. Furthermore, the wages in young and incumbent establishments are

analyzed more closely to test whether startups provide additional benefits to disadvantaged

workers in terms of lower wage penalties.

5.1 EMPLOYMENT OPPORTUNITIES IN STARTUPS

We estimate the probability of being hired by a startup as opposed to an incumbent

establishment using a linear probability model.16 In our first specification, the binary

dependent variable indicates whether the establishment is young, i.e., in its first 5 years of

business, or incumbent, i.e., older than 5 years (columns 1 and 2 in Table 4). Secondly, we

15 To identify workers who lost their jobs due to plant closures, we first identified closures in the

BHP making use of worker flow information by Hethey-Maier and Schmieder (2013) – similar to

the identification of startups – in order to distinguish between “true” closures and ID changes. In

a second step, we identified those workers who left closing plants in the same year in which they

closed down.

16 Linear probability models, i.e., ordinary least squares estimations with binary dependent

variables, produce consistent and unbiased estimates of partial effects (see, e.g., Wooldridge

2010, p. 562). As a robustness check, we also estimate a probit model. Average marginal effects

show the same level of significance and their magnitude is very similar to the coefficients

obtained in the linear probability model.

Page 14: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

12

test whether results are similar when we consider only new establishments in their first year

of business and compare them to the control group of incumbents (columns 3 and 4). Each

specification is estimated for two samples. The first includes all hires (columns 1 and 3),

allowing us to analyze the hiring behavior of startups and incumbents more generally and to

investigate whether startups are more or less likely to hire labor market entrants. The second

and more restrictive sample (columns 2 and 4) is limited to persons with previous labor

market experience, which enables us to include information on the previous employer as

additional explanatory variables.17

(Table 4 about here)

Table 4 presents the results of these estimations. The lower panel of this table indicates that

most of our control variables have statistically significant relationships with the probability

to be hired in a young or new plant rather than an incumbent plant. Mainly focusing on the

potentially employment-inhibiting characteristics of disadvantaged workers, we see from the

results in Table 4 that the probability to be hired by a young plant is 0.94 percentage points

higher for the oldest group of workers aged above 50 (compared to the middle-aged reference

group) in the sample for all hires (column 1). This effect is comparably small and not in all

samples statistically significant. In contrast to older workers, young workers are less likely

to be hired by a startup than the middle-aged reference group.18 Compared to workers with

German nationality, foreigners are considerably more likely to be hired by a young plant.

The coefficient of 0.0509 in the sample of all hires (column 1) indicates that the probability

of being hired by a young plant is ceteris paribus 5.09 percentage points higher for foreigners

than for workers with German nationality. In the other three estimations, the corresponding

coefficient estimates are slightly smaller but still positive and highly significant. In terms of

qualification, both low- and high-qualified workers are less likely to enter startups than the

reference group of medium-qualified workers, but for the disadvantaged group of low-

qualified workers, the estimated coefficients lose statistical significance in the more

restrictive sample. Included as an indicator for instable or perforated employment

biographies, the time of benefit receipt (relative to a person’s complete working life) also

has a positive impact on the likelihood of being hired by a startup in all estimations.

Employees coming from unemployment or from outside the labor force are more likely to

be recruited by young rather than incumbent establishments compared to workers coming

17 Additional to our main variables of interest, we include attributes of the new job as control

variables, namely indicators for vocational training, part-time work, and marginal employment.

Moreover, dummies for industry, year, and labor market region are included. Labor market

regions are classified according to Kropp and Schwengler (2011), who used workers’ commuting

patterns to define labor market regions.

18 The relationship with worker age is broadly in line with previous results by Coad et al. (2017) for

Denmark but in contrast with findings by Ouimet and Zarutskie (2014) for the U.S.

Page 15: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

13

directly from another employer.19 The size of these effects is between one and two

percentage points and thus not particularly large. When looking at new plants only in their

founding year, however, the positive effect for workers hired from unemployment vanishes

or even becomes negative (columns 3 and 4). In both specifications, first-time entrants to the

labor market are significantly less likely to be hired by startups than workers with previous

labor market experience. A rather large effect is found for workers affected by a closure of

their previous workplace (this information is only available in the restricted sample): the

estimated coefficients indicate a 12.29 (15.20) percentage points higher likelihood of

entering a young (new) establishment.

As startups are typically small, it could be argued that the employee sorting patterns found

may occur on establishment size rather than age (see also Burton et al. 2017). However,

comparing small startups with small incumbents may be misleading: For startups, being

small is not per se a negative attribute as they may have good prospects for future growth.

For incumbent plants, a small number of employees indicates that they have hardly grown

over time (e.g., because they are not very successful and possess further unfavorable

characteristics not observable in our data). Consequently, small incumbent plants might be

in an even more unfavorable position in the labor market than startups because they are

lacking credible growth prospects, so that incumbent plants of the same size do not constitute

an adequate comparison group for startups.

(Table 5 about here)

Despite this reservation, we also estimate our regressions for plants with maximum 20

employees to make startups and incumbents better comparable with respect to size. The

results, which are presented in Table 5, indicate that even conditional on being a small plant,

establishment age still plays an important role for employee sorting. Across all explanatory

variables (including controls), hiring patterns are largely in line with our main results

presented above. Specifically, young plants still have a higher probability than incumbents

to hire foreigners, workers with instable employment biographies and individuals from

outside the labor force. For foreigners and workers with instable employment biographies,

the effects are even stronger than in our main specification in Table 4. Young plants are still

more likely to hire employees affected by a closure of their last workplace and less likely to

hire first-time entrants to the labor market, and their hiring behavior towards other groups of

employees, such as females and young workers, remains in line with the main results. That

said, the results on older, low-qualified and unemployed workers are not robust in that they

flip signs. Focusing on new versus incumbent plants, most hiring patterns are also quite

similar to those presented in Table 4. These insights also hold if we further restrict the sample

19 We allow for a maximum gap of three months of non-employment. Results remain robust when

reducing this gap to one month.

Page 16: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

14

to plants with not more than 10 employees (results are available on request). Taken together,

we may conclude that even in a subsample of small plants, young plants are more likely than

incumbents to provide employment opportunities for various groups of disadvantaged

workers.

To test whether the results remain robust when looking at different sectors and workers, we

re-estimate the regression for several subgroups. The coefficients of our main variables of

interest are presented in Table 6. The estimates of the basic regression correspond to

specification 2 in Table 4, i.e., young vs. incumbent plants for workers with previous labor

market experience. Firstly, establishments in the secondary and tertiary sector are

investigated separately to test whether the results are driven by only one segment of the

economy (columns 2 and 3). Secondly, we distinguish between male and female hires

(columns 4 and 5). Estimates are to a large extent robust among the different subgroups.

Startups are more likely to hire workers with foreign nationality (except for females), with

relatively more time in benefit receipt, previous non-employment, or closure of the prior

workplace. Only the effect for older workers is mostly insignificant and the coefficient for

low qualification varies in sign and significance.

(Table 6 about here)

Coming back to our theoretical considerations on employment in startups outlined above,

most of the expected relationships sketched in Table 1 are confirmed by the regression

analyses. Indeed, workers with foreign nationality or instable employment biographies have

a higher probability of entering a newly founded firm, as startups might find it difficult to

attract other applicants and disadvantaged workers will have problems finding a job with a

less risky employer. That startups are more likely to hire foreigners confirms findings for

Sweden by Nyström (2012), and that instable employment biographies come along with an

increased probability to be hired by a startup corresponds with previous results for Germany

by Schnabel et al. (2011). Moreover, young firms seem to be less able to poach workers from

other employers and hence recruit individuals from outside the labor force or from

unemployment, which is generally in line with results for Denmark by Coad et al. (2017).20

The finding that the latter effect vanishes for new firms in their very first year, putting the

respective finding by Coad et al. (2017) into question, can potentially be explained if very

new firms regard the quality and expertise of their initial employees hired in the critical early

stage as too crucial to rely on workers whose current spell of unemployment can be seen as

20 Workers classified as hired from outside the labor force in our analyses may also have been

previously self-employed, which we cannot observe in our data and which could affect the way

in which our result has to be interpreted. However, it also appears quite sensible that previously

self-employed workers are more likely to enter startups than workers coming from dependent

employment, since preferences for entrepreneurial job characteristics are found to be strongly

pronounced among both, individuals founding their own business and workers joining a startup

(Roach and Sauermann 2015).

Page 17: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

15

a negative signal. The conjecture that startups, which are lacking established work routines

and are therefore more reliant on their employees’ expertise, might value work experience

more than incumbents is corroborated by the finding that they are less likely to hire labor

market entrants. This is in accordance with findings for Denmark by Coad et al. (2017) but

stands in contrast to insights for Sweden by Nyström (2012). The very strong effect of

previous employer’s closure on individuals’ probability of being hired by a startup, which

confirms results for Sweden by Nyström and Elvung (2015), suggests that young firms might

indeed help to overcome the negative consequences of reallocation and structural change.

5.2 WAGES IN STARTUPS

To investigate whether unfavorable worker characteristics are more or less strongly

penalized in startups than in incumbents, we estimate Mincer-type regressions on the entry

wages of new hires. Since our data set lacks information on working hours, we restrict our

analysis to regular full-time employees, thereby excluding part-time and marginal

employees as well as apprentices.21 As the dependent variable, we use the logarithm of daily

wages deflated by the consumer price index (in 2010 prices). To exclude implausibly low

wages, we delete observations in the lowest percentile of the wage distribution. Since in the

IEB data set wages are reported only up to the contribution limit to social security, we impute

higher wages in order to obtain unbiased estimates in the line of Gartner (2005).22

Regressions are run separately for men (columns 1 and 2 in Table 7) and women (columns

3 and 4).

In a first basic specification, we test whether the worker characteristics of interest are

associated with wage penalties in general (columns 1 and 3). In a second specification, we

interact these characteristics with a dummy indicating whether workers were hired by a

startup or by an incumbent to test whether wage penalties are stronger in young or incumbent

plants (columns 2 and 4). Additionally, the usual controls for work experience, establishment

size, occupational groups according to the categorization by Blossfeld (1987), industry, labor

market region, and year are included.

(Table 7 about here)

21 Moreover, we focus on hires with previous work experience. This is done because work

experience is included as a control variable in the wage regressions and since first time entrants

to the labor market are characterized by zero work experience, we cannot separately identify a

wage effect for this group of workers by including a dummy variable. As we do not want to mix

up job-to-job transitions and first-time labor market entrants in the reference category in our

regression analyses, we decided to exclude the latter.

22 When estimating the regression using only non-censored wages below the social security

contribution threshold, our insights remain unchanged; results are available on request.

Page 18: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

16

Table 7 presents the results for the Mincer wage regressions. The control variables in all

regressions are statistically significant and show the expected signs. Coming to our main

variables of interest, it can be seen that for men the estimated coefficient for entering a young

establishment as opposed to an incumbent is significantly negative. Ceteris paribus, wages

in startups are 3.44 percent lower than in incumbents in the basic specification (column 1).23

As expected, old age, foreign nationality, low qualification, having spent a large fraction of

one’s working life in benefit receipt, as well as recruitment from unemployment or from

outside the labor force are all associated with significantly lower wages.

In the second specification (column 2), interaction terms between these characteristics and

the startup indicator allow for a more detailed analysis of wage penalties. Our results suggest

that the negative relation between unfavorable worker characteristics and wages is even

more pronounced for those workers hired by startups. Being older than 50 years is associated

with 6.99 percent lower wages in incumbents and beyond that, older workers hired by

startups face an additional wage penalty of 2.31 percent. The difference in wage penalties

between startups and incumbents is even larger for foreign employees, amounting to 8.69

percent. The wage penalty for foreigners is therefore more than five times higher in startups

than in incumbents. Being low-qualified, having spent a relatively large fraction of time in

benefit receipt, and coming from unemployment or from outside the labor market also comes

along with higher wage penalties in startups than in incumbent plants. For workers coming

from outside the labor force, the wage penalty in startups is almost three times higher than

in incumbents whereas the differences for the other groups are comparably small. Additional

regressions conducted as robustness checks but not reported show that these insights still

hold when focusing on new establishments in their first year of business (rather than young

establishments in their first five years) and comparing them to incumbents.

Estimating the model for women, the basic specification (column 3) suggests that wages in

startups are on average slightly higher than in incumbents, though the difference is hardly

significant. As was the case for men, disadvantaged groups of female workers earn lower

wages than their counterparts with more favorable attributes. Only the estimate for foreign

nationality is statistically insignificant. In the second specification (column 4), the inclusion

of interaction terms shows that also female workers who are foreign, low-qualified, or

coming from outside the labor force are penalized more strongly in startups but overall the

differences between startups and incumbents are less pronounced than for men. The

interaction term is only positive for older female workers above 50 and it is statistically

23 These results are in line with previous studies finding significantly lower wages in newly founded

firms. For Germany, Brixy et al. (2007) estimate a stronger effect of 8 percent; however, they can

only compare mean wages between establishments and control for the average workforce

composition. Nyström and Elvung (2014) for Sweden focus on labor market entrants and find a

wage penalty of approximately 2.9 percent for entering a startup after matching on worker

characteristics.

Page 19: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

17

insignificant for women who are recruited from unemployment or who received

unemployment benefits for a large fraction of their working life. Again, our robustness check

of focusing on new establishments renders the same insights.

Our results thus do not suggest that the employment opportunities which startups provide to

disadvantaged workers come along with additional benefits in terms of lower wage penalties

associated with unfavorable worker characteristics. Instead, wage penalties seem even more

pronounced in startups for all groups of disadvantaged employees among men, and for most

of these groups among women.24 Our findings are hence in line with the conjecture that

incumbents, which are more likely to have elaborate personnel regulations and wage setting

institutions such as works councils and collective bargaining agreements, have less scope

for paying lower wages to disadvantaged groups of new hires. It is, however, unclear whether

the higher wage penalties in startups reflect these firms’ ability to discriminate more against

these groups of workers. It could also be the case that wage setting institutions like collective

agreements or works councils lead to a compression of the wage structure, i.e.,

disadvantaged workers might receive wages above their value of marginal product in

incumbent establishments, whereas wages for these groups of workers in startups reflect

their true productivity.

6. CONCLUSIONS

Although startups currently receive strong political attention and their contribution to

economic growth and job creation has been analyzed by a wide range of studies, much less

is known about their hiring behavior and the quality of employment created by these young

firms. To shed some light on these aspects, we have investigated whether startups are more

likely than incumbents to provide employment opportunities for disadvantaged workers or

for persons who lost their previous job in the course of structural change, thereby helping

these individuals to stay attached to the labor market and prevent human capital

depreciations. Moreover, we have analyzed whether wage penalties associated with

unfavorable worker characteristics are more or less pronounced in startups than in

incumbents.

24 Focusing on entry wages, we are not able to control for worker fixed effects, so that we cannot

fully rule out that particular groups of disadvantaged workers who enter startups or incumbents,

respectively, still differ to some extent in unobserved characteristics (e.g., foreigners in their

command of the German language, which also might affect their chance of being hired by an

incumbent plant). That said, as our wage regressions include very detailed information about

workers' employment biographies and other characteristics, we believe that this potential bias is

not large.

Page 20: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

18

Our results based on a large linked employer-employee data set for Germany suggest that

startups are indeed more likely to hire several groups of disadvantaged workers than

incumbent firms. We find that non-German applicants and workers with instable

employment biographies have a higher likelihood of being hired by a young establishment.

The same applies to workers recruited from unemployment or outside the labor force and to

workers who were affected by a plant closure. These findings potentially reflect that startups

– due to their lack of reputation and their perceived riskiness – find it difficult to attract other

applicants and that disadvantaged workers may have problems finding jobs with less risky

employers. Focusing on the monetary quality of employment in startups, our analysis of

entry wages in young and incumbent establishments indicates that the wage penalties

associated with disadvantageous worker characteristics are more pronounced in young

establishments, probably due to a lack of elaborate wage setting institutions. Hence, startups

seem to be either able to discriminate more against certain groups of workers or to pay wages

that are more closely related to worker productivity. To some extent, the finding that startups

are able to pay lower wages to disadvantaged groups of workers could also explain why

young plants are more likely than incumbents to hire them.

When interpreting our main finding that startups are more likely than incumbents to provide

employment opportunities to disadvantaged workers, one should keep in mind that it does

not allow us to draw positive conclusions on the question whether startups and their

subsidization can be regarded as beneficial for these groups of workers. Firstly, in terms of

entry wages we find job conditions for disadvantaged workers to be even worse in startups

than in incumbents. Secondly, supporting job creation in incumbent establishments might –

at least indirectly – have the same consequences as the subsidization of startups: If

established plants themselves are perceived as more attractive employers and thus are able

to fill the newly created positions with the most desired workers poached from other firms,

this might trigger a redeployment process which ultimately results in the least attractive

employers hiring from the groups of disadvantaged workers. Thus, while facilitating the

foundation and growth of startups might be beneficial for the entire economy by fostering

competition and growth, the skepticism towards the advantages of startup subsidization in

terms of job creation uttered in some studies (e.g., Santarelli and Vivarelli 2007; Shane 2009)

cannot be curbed by our analysis. Finally, whereas the empirical analysis in this paper has

focused on workers’ entry wages, it would be of further interest to investigate whether

entering startups rather than incumbents turns out to be beneficial in the long run. Future

research should therefore analyze wage trajectories and subsequent career paths of workers

hired by startups, thereby taking establishment survival and employees’ voluntary and

involuntary separations into account.

Page 21: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

19

REFERENCES

Antoni, Manfred, Andreas Ganzer, and Phillipp vom Berge. 2016. Sample of Integrated

Labour Market Biographies (SIAB) 1975-2014. FDZ-Datenreport No. 04/2016. Nürnberg:

Institute for Employment Research.

Audretsch, David B., George van Leeuwen, Bert Menkveld, and Roy Thurik. 2001. Market

dynamics in the Netherlands: Competition policy and the role of small firms. International

Journal of Industrial Organization 19(5): 795-821.

Blossfeld, Hans-Peter. 1987. Labor market entry and the sexual segregation of careers in the

Federal Republic of Germany. American Journal of Sociology 93(1): 89-118.

Brixy, Udo, Susanne Kohaut, and Claus Schnabel. 2007. Do newly founded firms pay lower

wages? First evidence from Germany. Small Business Economics 29(1-2): 161-71.

Brown, Charles, and James L. Medoff. 2003. Firm Age and Wages. Journal of Labor

Economics 21(3): 677-97.

Brüderl, Josef, Peter Preisendörfer, and Rolf Ziegler. 2007. Der Erfolg neugegründeter

Betriebe. 3rd ed. Berlin: Duncker & Humblot.

Bundesagentur für Arbeit. 2016. Arbeitsmarkt 2015. Nürnberg: Bundesagentur für Arbeit.

Burton, M. Diane, Michael S. Dahl, and Olav Sorenson. 2017. Do Start-ups Pay Less?

Industrial and Labor Relations Review, forthcoming (https://doi.org/10.1177/

0019793917747240).

Coad, Alex, Kristian Nielsen, and Bram Timmermans. 2017. My first employee: an

empirical investigation. Small Business Economics 48(1): 25-45.

Dahl, Michael S., and Steven Klepper. 2015. Whom do new firms hire? Industrial and

Corporate Change 24(4): 819-36.

Dahl, Michael S, and Toke Reichstein. 2007. Are you experienced? Prior experience and the

survival of new organizations. Industry and Innovation 14(5): 497-511.

Dinlersoz, Emin M., Henry R. Hyatt, and Hubert P. Janicki. 2016. Who Works for Whom?

Worker Sorting in a Model of Entrepreneurship with Heterogeneous Labor Markets. IZA

Discussion Paper No. 9693. Bonn: Institute for the Study of Labor.

Ellguth, Peter, Susanne Kohaut, and Iris Möller. 2014. The IAB Establishment Panel –

methodological essentials and data quality. Journal for Labour Market Research 47(1-2):

27-41.

Page 22: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

20

Fackler, Daniel, and Eva Hank. 2016. Who buffers income losses after job displacement?

The role of alternative income sources, the family, and the state. SOEPpapers on

Multidisciplinary Panel Data Research No. 863. Berlin: German Institute for Economic

Research.

Fackler, Daniel, Claus Schnabel, and Joachim Wagner. 2013. Establishments exits in

Germany: the role of size and age. Small Business Economics 41(3): 683-700.

Fairlie, Robert W., and Javier Miranda. 2017. Taking the Leap: The Determinants of

Entrepreneurs Hiring their First Employee. Journal of Economics & Management Strategy

26(1): 3-34.

Fitzenberger, Bernd, Aderonke Osikominu, and Robert Völter. 2005. Imputation rules to

improve the education variable in the IAB employment subsample. FDZ-Methodenreport

No. 03/2005. Nürnberg: Institute for Employment Research.

Fritsch, Michael, and Udo Brixy. 2004. The establishment file of the German social

insurance statistics. Schmollers Jahrbuch 124(1): 183-90.

Fritsch, Michael, and Florian Noseleit. 2013. Investigating the anatomy of the employment

effect of new business formation. Cambridge Journal of Economics 37(2): 349-77.

Fritsch, Michael, and Antje Weyh. 2006. How Large are the Direct Employment Effects of

New Businesses? An Empirical Investigation for West Germany. Small Business Economics

27(2-3): 245-60.

Fuchs, Michaela, and Antje Weyh. 2010. The determinants of job creation and destruction:

plant-level evidence for Eastern and Western Germany. Empirica 37(4): 425-44.

Gartner, Hermann. 2005. The imputation of wages above the contribution limit with the

German IAB employment sample. FDZ-Methodenreport No. 02/2005. Nürnberg: Institute

for Employment Research.

Geroski, Paul A. 1995. What do we know about entry? International Journal of Industrial

Organization 13(4): 421-40.

Geroski, Paul A., José Mata, and Pedro Portugal. 2010. Founding conditions and the survival

of new firms. Strategic Management Journal 31(5): 510-29.

Haltiwanger, John C., Ron S. Jarmin, and Javier Miranda. 2013. Who creates jobs? Small

versus large versus young. Review of Economics and Statistics 95(2): 347-61.

Page 23: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

21

Hethey-Maier, Tanja, and Johannes F. Schmieder. 2013. Does the use of worker flows

improve the analysis of establishment turnover? Evidence from German administrative data.

Schmollers Jahrbuch 133(4): 477-510.

Kaas, Leo, and Christian Manger. 2012. Ethnic Discrimination in Germany's Labour Market:

A Field Experiment. German Economic Review 13(1): 1-20.

Koch, Andreas, Jochen Späth, and Harald Strotmann. 2013. The role of employees for post-

entry firm growth. Small Business Economics 41(3): 733-55.

Kropp, Per, and Barbara Schwengler. 2011. Abgrenzung von Arbeitsmarktregionen – ein

Methodenvorschlag. Raumforschung und Raumordnung 69(1): 45-62.

Michelacci, Claudio, and Vincenzo Quadrini. 2005. Borrowing from employees: Wage

dynamics with financial constraints. Journal of the European Economic Association 3(2-3):

360-69.

Möller, Joachim, and Ulrich Walwei (Eds.). 2017. Arbeitsmarkt kompakt - Analysen, Daten,

Fakten. Bielefeld: Bertelsmann.

Nyström, Kristina. 2012. Labor mobility and entrepreneurship: who do new firms employ?

In Charlie Karlsson, Börje Johansson, and Roger R. Stough (Eds.), Entrepreneurship, Social

Capital and Governance. Directions for the Sustainable Development and Competitiveness

of Regions, pp 102-14. Cheltenham, Northampton: Edward Elgar.

Nyström, Kristina, and Gulzat Zhetibaeva Elvung. 2014. New firms and labor market

entrants: Is there a wage penalty for employment in new firms? Small Business Economics

43(2): 399-410.

Nyström, Kristina, and Gulzat Zhetibaeva Elvung. 2015. New Firms as Employers: The

Wage Penalty for Voluntary and Involuntary Job Switchers. LABOUR 29(4): 348-66.

Ouimet, Paige, and Rebecca Zarutskie. 2014. Who works for startups? The relation between

firm age, employee age, and growth. Journal of Financial Economics 112(3): 386-407.

Roach, Michael, and Henry Sauermann. 2015. Founder or Joiner? The Role of Preferences

and Context in Shaping Different Entrepreneurial Interests. Management Science 61(9):

2160-84.

Rocha, Vera, Mirjam C. van Praag, Timothy B. Folta, and Anabela Carneiro. 2016.

Entrepreneurial Choices of Initial Human Capital Endowments and New Venture Success.

IZA Discussion Paper No. 9919. Bonn: Institute for the Study of Labor.

Page 24: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

22

Santarelli, Enrico, and Marco Vivarelli. 2007. Entrepreneurship and the process of firms’

entry, survival and growth. Industrial and Corporate Change 16(3): 455-88.

Schmieder, Johannes F. 2013. What causes wage dispersion? Evidence from new firms.

Working Paper. Boston University. Available at: https://sites.google.com/site

/johannesschmieder/Schmieder_WagesNewFirms_WorkingPaper.pdf?attredirects=0&d=1)

.

Schmucker, Alexandra, Stefan Seth, Johannes Ludsteck, Johanna Eberele, and Andreas

Ganzer. 2016. Establishment History Panel 1975-2014. FDZ Datenreport No. 03/2016,

Nürnberg: Institute for Employment Research.

Schnabel, Claus, Susanne Kohaut, and Udo Brixy. 2011. Employment stability in newly

founded firms: a matching approach using linked employer-employee data from Germany.

Small Business Economics 36(1): 85-100.

Shane, Scott. 2009. Why encouraging more people to become entrepreneurs is bad public

policy. Small Business Economics 33(2): 141-49.

Wagner, Joachim. 2006. Firmenalter und Firmenperformance: Empirische Befunde zu

Unterschieden zwischen jungen und alten Firmen in Deutschland. In Lutz Bellmann and

Joachim Wagner (Eds.), Betriebsdemographie, pp. 83-111. Nürnberg: Institut für

Arbeitsmarkt- und Berufsforschung.

Weber, Andrea, and Christine Zulehner. 2010. Female Hires and the Success of Start-up

Firms. American Economic Review (Papers & Proceedings) 100(2): 358-61.

Weber, Andrea, and Christine Zulehner. 2014. Competition and gender prejudice: Are

discriminatory employers doomed to fail? Journal of the European Economic Association

12(2): 492-521.

Wooldridge, Jeffrey M. 2010. Econometric Analysis of Cross Section and Panel Data. 2nd

ed. Cambridge, MA: MIT Press.

Page 25: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

23

TABLES

Table 1: Selected worker characteristics and their expected relationship with the probability

of joining newly founded firms

Worker characteristics Relationship with employment in startups

Firm perspective Worker perspective

Older worker (age above 50 years) + +

Foreign nationality + +

Low-qualified +/- +

Instable employment biography +/- +

Recruited from unemployment + +

Recruited from out of the labor force + +/-

First-time entrant to the labor market +/- -

Last establishment: closure + +

Table 2: Selected establishment characteristics of young, new, and incumbent plants

(means)

Young Plants New Plants Incumbent

Plants

Number of Employees 5.29 (15.20)

2.85 (2.84)

31.17 (281.96)

Secondary Sector (%) 19.28 (39.45)

18.54 (38.86)

30.54 (46.06)

Tertiary Sector (%) 80.72 (39.45)

81.46 (38.86)

69.46 (46.06)

Share of Women (%) 49.37 (39.01)

48.31 (42.66)

50.01 (32.58)

Share of Full-Time Workers (%) 48.21 (38.10)

51.97 (41.91)

50.90 (31.51)

Share of Marginally Employed

Workers (%) 29.10 (30.96)

22.54 (29.41)

29.57 (29.82)

Number of Observations 221,736 94,704 294,770

Notes: The sample includes only West German establishments in the years 2003-2014, excluding agriculture, energy &

mining, and the public and non-profit sector. Standard deviations are reported in parentheses. Source: BHP, own

calculations.

Page 26: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

24

Table 3: Selected individual characteristics of workers entering young, new, and incumbent

plants (means)

Young Plants New Plants Incumbent

Plants

Age: Up to 30 years (%) 46.69 (49.89)

40.84 (49.15)

54.27 (49.82)

Age: 31-50 years (%) 44.13 (49.65)

47.97 (49.96)

37.98 (48.53)

Age: Above 50 years (%) 9.18 (28.88)

11.18 (31.52)

7.75 (26.74)

Foreign Nationality (%) 13.81 (34.50)

14.57 (35.28)

10.20 (30.26)

Women (%) 47.77 (49.95)

49.89 (50.00)

44.40 (49.69)

Low-Qualified (%) 21.05 (40.77)

19.01 (39.23)

24.65 (43.09)

Medium-Qualified (%) 69.95 (45.85)

72.62 (44.59)

63.29 (48.20)

High-Qualified (%) 9.00 (28.62)

8.38 (27.70)

12.06 (32.57)

Years since first appearance in the data 14.48 (10.52)

15.87 (10.64)

12.81 (10.71)

…thereof time of benefit receipt (%) 8.83 (14.23)

8.61 (13.54)

6.82 (13.20)

Origin: Job-to-Job Transition (%) 56.55 (49.57)

58.72 (49.23)

57.43 (49.45)

Origin: From Unemployment (%) 21.73 (41.24)

20.59 (40.43)

17.04 (37.60)

Origin: From Outside the Labor

Force (%) 17.24 (37.77)

17.71 (38.17)

15.90 (36.56)

Origin: First-Time Entrant (%) 4.48 (20.68)

2.99 (17.02)

9.63 (29.50)

Number of Previous Employers 6.43 (4.76)

6.65 (4.64)

5.55 (4.52)

Apprenticeship / Vocational Training (%) 3.80 (19.11)

2.97 (16.96)

8.09 (27.27)

Part-Time (%) 40.87 (49.16)

41.71 (49.31)

34.58 (47.56)

Marginally Employed (%) 28.38 (45.08)

25.54 (43.61)

24.49 (43.01)

Number of Observations 570,752 201,150 1,442,996

Last Establishment: Closure (%) 15.00 (35.71)

21.98 (41.41)

7.58 (26.47)

Last Establishment: Young (%) 31.47 (46.44)

33.57 (47.22)

22.63 (41.84)

Number of Observations 508,379 184,282 1,192,590

Notes: The sample includes only West German establishments in the years 2003-2014, excluding agriculture, energy &

mining, and the public and non-profit sector. Only individuals newly entering an establishment, excluding those older than

30 when first appearing in the IEB. Standard deviations are reported in parentheses. Source: IEB, BHP; own calculations.

Page 27: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

25

Table 4: Probability of being hired by a startup (young / new establishment) as opposed to an incumbent

YOUNG VS. INCUMBENT PLANTS NEW VS. INCUMBENT PLANTS

All Hires

Hires with Lab. Market

Experience All Hires

Hires with Lab. Market

Experience

(1) (2) (3) (4)

Age: Up to 30 years (d) -0.0300 (0.0035)

**

* -0.0228 (0.0029)

**

* -0.0375 (0.0016)

**

* -0.0311 (0.0015)

**

*

Age: 31-50 years (d) Reference Reference

Age: Above 50 years(d) 0.0094 (0.0027)

**

* 0.0036 (0.0023) 0.0226 (0.0015)

**

* 0.0169 (0.0014)

**

*

Foreign Nationality (d) 0.0509 (0.0047)

**

* 0.0368 (0.0044)

**

* 0.0383 (0.0027)

**

* 0.0318 (0.0027)

**

*

Low-Qualified (d) -0.0160 (0.0036)

**

* -0.0026 (0.0037) -0.0077 (0.0017)

**

* -0.0005 (0.0019)

Medium-Qualified (d) Reference Reference

High-Qualified (d) -0.0301 (0.0058)

**

* -0.0197 (0.0055)

**

* -0.0198 (0.0026)

**

* -0.0085 (0.0024)

**

*

Rel. Time of Benefit Receipt 0.0762 (0.0076)

**

* 0.1078 (0.0094)

**

* 0.0398 (0.0038)

**

* 0.0614 (0.0048)

**

*

Origin: Job-to-Job Transition (d) Reference Reference

Origin: From Unemployment (d) 0.0136 (0.0028)

**

* 0.0144 (0.0025)

**

* -0.0021 (0.0013) * -0.0012 (0.0012)

Origin: From Outside the Lab. Force (d) 0.0145 (0.0026)

**

* 0.0231 (0.0022)

**

* 0.0135 (0.0012)

**

* 0.0219 (0.0013)

**

*

Origin: First-Time Entrant (d) -0.0544 (0.0042)

**

* -0.0293 (0.0021)

**

*

Last Establishment: Closure (d) 0.1229 (0.0042)

**

* 0.1520 (0.0026)

**

*

Female (d) -0.0063 (0.0032) ** -0.0077 (0.0031) ** 0.0017 (0.0015) 0.0011 (0.0015)

Vocational Training (d) -0.0863 (0.0057)

**

* -0.0878 (0.0054)

**

* -0.0546 (0.0029)

**

* -0.0571 (0.0027)

**

*

Part-Time (d) 0.0112 (0.0066) * 0.0120 (0.0062) * 0.0239 (0.0037)

**

* 0.0237 (0.0035)

**

*

Marginally Employed (d) -0.0197 (0.0063)

**

* -0.0203 (0.0059)

**

* -0.0527 (0.0039)

**

* -0.0542 (0.0037)

**

*

Number of Previous Employers 0.0032 (0.0004)

**

* 0.0023 (0.0003)

**

* 0.0017 (0.0002)

**

* 0.0009 (0.0001)

**

*

Page 28: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

26

Last Establishment: Young (d) 0.0522 (0.0022)

**

* 0.0313 (0.0012)

**

*

Last Establishment: Industry (2-Digit) (d) Included

**

* Included

**

*

Industry (2-Digit) (d) Included

**

* Included

**

* Included

**

* Included

**

*

Labor Market Region (d) Included

**

* Included ** Included

**

* Included

**

*

Year (d) Included

**

* Included

**

* Included

**

* Included

**

*

Constant 0.1637 (0.0221)

**

* 0.1328 (0.0200)

**

* 0.0796 (0.0091)

**

* 0.0533 (0.0097)

**

*

Number of Observations 2,013,748 1,700,969 1,644,146 1,376,872

R² 0.0948 0.1026 0.0767 0.0981

Notes: OLS regressions. The binary dependent variable indicates whether an individual is newly hired in a young/new (1) or incumbent (0) establishment. Further sample restrictions as in

Table 3. Standard errors (reported in parentheses) are clustered by establishment. (d) denotes a dummy variable. */**/*** indicates statistical significance at the10/5/1% level, respectively.

Source: IEB, BHP, own calculations.

Page 29: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

27

Table 5: Probability of being hired by a startup (young/new establishment) for establishments with max. 20 employees

YOUNG VS. INCUMBENT PLANTS NEW VS. INCUMBENT PLANTS

All Hires

Hires with Lab. Market

Experience All Hires

Hires with Lab. Market

Experience

(1) (2) (3) (4)

Age: Up to 30 years (d) -0.0085 (0.0018)

**

* -0.0033 (0.0018) * -0.0460 (0.0019)

**

* -0.0336 (0.0019)

**

*

Age: 31-50 years (d) Reference Reference

Age: Above 50 years(d) -0.0124 (0.0022)

**

* -0.0177 (0.0022)

**

* 0.0209 (0.0024)

**

* 0.0098 (0.0023)

**

*

Foreign Nationality (d) 0.1081 (0.0030)

**

* 0.0884 (0.0030)

**

* 0.1125 (0.0034)

**

* 0.0955 (0.0035)

**

*

Low-Qualified (d) 0.0190 (0.0020)

**

* 0.0339 (0.0021)

**

* 0.0246 (0.0021)

**

* 0.0372 (0.0023)

**

*

Medium-Qualified (d) Reference Reference

High-Qualified (d) 0.0008 (0.0033) 0.0041 (0.0032) -0.0143 (0.0033)

**

* -0.0059 (0.0032) *

Rel. Time of Benefit Receipt 0.1033 (0.0054)

**

* 0.1308 (0.0067)

**

* 0.0920 (0.0059)

**

* 0.1271 (0.0074)

**

*

Origin: Job-to-Job Transition (d) Reference Reference

Origin: From Unemployment (d) -0.0083 (0.0017)

**

* -0.0032 (0.0017) * -0.0388 (0.0019)

**

* -0.0279 (0.0019)

**

*

Origin: From Outside the Lab. Force (d) 0.0095 (0.0017)

**

* 0.0236 (0.0018)

**

* -0.0040 (0.0019) ** 0.0180 (0.0020)

**

*

Origin: First-Time Entrant (d) -0.0954 (0.0030)

**

* -0.1034 (0.0028)

**

*

Last Establishment: Closure (d) 0.1212 (0.0021)

**

* 0.2039 (0.0025)

**

*

Female (d) -0.0242 (0.0022)

**

* -0.0233 (0.0022)

**

* -0.0265 (0.0021)

**

* -0.0246 (0.0021)

**

*

Vocational Training (d) -0.1893 (0.0035)

**

* -0.1757 (0.0042)

**

* -0.1907 (0.0030)

**

* -0.1857 (0.0037)

**

*

Part-Time (d) 0.0485 (0.0027)

**

* 0.0422 (0.0027)

**

* 0.0612 (0.0028)

**

* 0.0540 (0.0028)

**

*

Marginally Employed (d) -0.0787 (0.0028)

**

* -0.0800 (0.0028)

**

* -0.1500 (0.0029)

**

* -0.1487 (0.0029)

**

*

Number of Previous Employers 0.0046 (0.0002)

**

* 0.0038 (0.0002)

**

* 0.0036 (0.0002)

**

* 0.0029 (0.0002)

**

*

Page 30: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

28

Last Establishment: Young (d) 0.0533 (0.0015)

**

* 0.0508 (0.0017)

**

*

Last Establishment: Industry (2-Digit) (d) Included

**

* Included

**

*

Industry (2-Digit) (d) Included

**

* Included

**

* Included

**

* Included

**

*

Labor Market Region (d) Included

**

* Included

**

* Included

**

* Included

**

*

Year (d) Included

**

* Included

**

* Included

**

* Included

**

*

Constant 0.3571 (0.0143)

**

* 0.3017 (0.0165)

**

* 0.2615 (0.0108)

**

* 0.1750 (0.0136)

**

*

Number of Observations 868,303 749,374 609,804 524,964

R² 0.0612 0.0613 0.0830 0.0966

Notes: OLS regressions. The binary dependent variable indicates whether an individual is newly hired in a young/new (1) or incumbent (0) establishment. Further sample restrictions as in

Table 3. Standard errors (reported in parentheses) are clustered by establishment. (d) denotes a dummy variable. */**/*** indicates statistical significance at the10/5/1% level, respectively.

Source: IEB, BHP, own calculations.

Page 31: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

29

Table 6: Probability of being hired by a startup (young / new establishment) as opposed to an incumbent, estimates for different sectors and

subgroups of workers

YOUNG VS. INCUMBENT PLANTS

Hires with Labor Market Experience

Basic Regression

Establishments in

Secondary Sector

Establishments in

Tertiary Sector Male Hires Female Hires

(1) (2) (3) (4) (5)

Age: Above 50 years (d) 0.0036 -0.0012 0.0039 -0.0005 0.0058 **

(0.0023) (0.0030) (0.0029) (0.0029) (0.0027)

Foreign Nationality (d) 0.0368 *** 0.0545 *** 0.0312 *** 0.0602 *** -0.0061

(0.0044) (0.0067) (0.0053) (0.0049) (0.0051)

Low-Qualified (d) -0.0026 0.0113 *** -0.0051 0.0147 *** -0.0167 ***

(0.0037) (0.0040) (0.0044) (0.0039) (0.0041)

Relative Time of Benefit Receipt 0.1078 *** 0.1558 *** 0.0974 *** 0.1296 *** 0.0600 ***

(0.0094) (0.0109) (0.0112) (0.0116) (0.0099)

Origin: From Unemployment (d) 0.0144 *** 0.0173 *** 0.0131 *** 0.0188 *** 0.0063 **

(0.0025) (0.0029) (0.0031) (0.0028) (0.0027)

Origin: From Outside the Labor Force (d) 0.0231 *** 0.0277 *** 0.0212 *** 0.0381 *** 0.0089 *** (0.0022) (0.0033) (0.0027) (0.0030) (0.0023)

Last Establishment: Closure (d) 0.1229 *** 0.1051 *** 0.1280 *** 0.1118 *** 0.1321 ***

(0.0042) (0.0051) (0.0053) (0.0051) (0.0041)

Number of Observations 1,700,969 453,590 1,247,379 926,736 774,233

R² 0.1026 0.1618 0.0687 0.1221 0.0884 Notes: OLS regressions. The binary dependent variable indicates whether an individual is newly hired in a young (1) or incumbent (0) establishment. All variables listed in Table 4 are included

in each regression. Further sample restrictions as in Table 3. Standard errors (reported in parentheses) are clustered by establishment. “Basic Regression” refers to regression (2) in Table

4. (d) denotes a dummy variable. */**/*** indicates statistical significance at the10/5/1% level, respectively. Source: IEB, BHP, own calculations.

Page 32: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

30

Table 7: Determinants of entry wages

Men Women

Basic

(1)

Interaction

(2)

Basic

(3)

Interaction

(4)

Young Plant (d) -0.0344 (0.0067)

**

* -0.0255 (0.0089)

**

* 0.0137 (0.0082) * -0.0178 (0.0132)

Age: Up to 30 years (d) -0.0634 (0.0032)

**

* -0.0796 (0.0041)

**

* 0.0309 (0.0054)

**

* 0.0058 (0.0071)

Age: 31-50 years (d) Reference Reference

Age: Above 50 years (d) -0.0795 (0.0031)

**

* -0.0699 (0.0037)

**

* -0.0983 (0.0043)

**

* -0.1026 (0.0055)

**

*

Age: Up to 30 * Young Plant 0.0501 (0.0063)

**

* 0.0796 (0.0101)

**

*

Age: Above 50 * Young Plant -0.0231 (0.0074)

**

* 0.0188 (0.0091) **

Foreign Nationality (d) -0.0501 (0.0032)

**

* -0.0189 (0.0037)

**

* -0.0046 (0.0045) 0.0019 (0.0060)

Foreign Nationality * Young Plant -0.0869 (0.0060)

**

* -0.0218 (0.0081)

**

*

Low-Qualified (d) -0.0906 (0.0047)

**

* -0.0842 (0.0066)

**

* -0.0831 (0.0055)

**

* -0.0742 (0.0069)

**

*

Medium-Qualified (d) Reference Reference

High-Qualified (d) 0.3355 (0.0050)

**

* 0.3252 (0.0058)

**

* 0.2957 (0.0072)

**

* 0.2922 (0.0089)

**

*

Low-Qualified * Young Plant -0.0176 (0.0077) ** -0.0307 (0.0085)

**

*

High-Qualified * Young Plant 0.0394 (0.0129)

**

* 0.0107 (0.0120)

Rel. Time of Benefit Receipt -0.2695 (0.0082)

**

* -0.2542 (0.0102)

**

* -0.2627 (0.0126)

**

* -0.2753 (0.0171)

**

*

Time of Benefit Receipt * Young Plant -0.0514 (0.0183)

**

* 0.0317 (0.0285)

Origin: Job-to-Job Transition (d) Reference Reference

Origin: From Unemployment (d) -0.1006 (0.0023)

**

* -0.0975 (0.0029)

**

* -0.0882 (0.0031)

**

* -0.0856 (0.0037)

**

*

Origin: From Outside the Lab. Force (d) -0.1122 (0.0051)

**

* -0.0681 (0.0081)

**

* -0.1422 (0.0070)

**

* -0.1214 (0.0100)

**

*

Page 33: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

31

Origin: From Unempl. * Young Plant -0.0124 (0.0048) ** -0.0092 (0.0069)

Origin: From Outside * Young Plant -0.1304 (0.0101)

**

* -0.0579 (0.0132)

**

*

Number of Previous Employers -0.0108 (0.0003)

**

* -0.0108 (0.0003)

**

* -0.0061 (0.0006)

**

* -0.0061 (0.0006)

**

*

Work Experience 0.0440 (0.0012)

**

* 0.0444 (0.0011)

**

* 0.0391 (0.0016)

**

* 0.0390 (0.0015)

**

*

Work Experience² -0.0016 (0.0001)

**

* -0.0016 (0.0001)

**

* -0.0017 (0.0001)

**

* -0.0018 (0.0001)

**

*

Work Experience³ 0.0000 (0.0000)

**

* 0.0000 (0.0000)

**

* 0.0000 (0.0000)

**

* 0.0000 (0.0000)

**

*

Establishment Size:1-4 (d) -0.3855 (0.0161)

**

* -0.3827 (0.0160)

**

* -0.5103 (0.0293)

**

* -0.5085 (0.0293)

**

*

Establishment Size:5-9 (d) -0.2956 (0.0153)

**

* -0.2960 (0.0152)

**

* -0.3768 (0.0280)

**

* -0.3781 (0.0279)

**

*

Establishment Size:10-19 (d) -0.2438 (0.0150)

**

* -0.2447 (0.0149)

**

* -0.3077 (0.0278)

**

* -0.3091 (0.0276)

**

*

Establishment Size:20-49 (d) -0.2004 (0.0149)

**

* -0.2013 (0.0148)

**

* -0.2360 (0.0275)

**

* -0.2371 (0.0273)

**

*

Establishment Size:50-99 (d) -0.1872 (0.0154)

**

* -0.1876 (0.0152)

**

* -0.1869 (0.0278)

**

* -0.1879 (0.0277)

**

*

Establishment Size:100-199 (d) -0.1730 (0.0161)

**

* -0.1730 (0.0159)

**

* -0.1580 (0.0287)

**

* -0.1586 (0.0286)

**

*

Establishment Size: 200-499 (d) -0.1148 (0.0157)

**

* -0.1148 (0.0156)

**

* -0.1138 (0.0272)

**

* -0.1142 (0.0271)

**

*

Establishment Size: at least 500 (d) Reference Reference

Occupation (Blossfeld) (d) Included

**

* Included

**

* Included

**

* Included

**

*

Industry (2-Digit) (d) Included

**

* Included

**

* Included

**

* Included

**

*

Labor Market Region (d) Included

**

* Included

**

* Included

**

* Included

**

*

Year (d) Included

**

* Included

**

* Included

**

* Included

**

*

Constant 4.3000 (0.0247)

**

* 4.2973 (0.0245)

**

* 3.9188 (0.0404)

**

* 3.9293 (0.0395)

**

*

Number of Observations 686,992 686,992 312,005 312,005

R² 0.6018 0.6038 0.4579 0.4591

Page 34: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and

32

Notes: OLS regressions. Dependent variable is the logarithm of daily wages in 2010 Euros, where those wages above the contribution limit to social security are imputed. Wages in the lowest

percentile of the distribution are excluded. Only regular full-time employees with previous work experience. Further sample restrictions as in Table 3. Standard errors (reported in parentheses)

are clustered by establishment. (d) denotes a dummy variable. */**/*** indicates statistical significance at the10/5/1% level, respectively. Source: IEB, BHP, own calculations.

Page 35: Do startups provide employment opportunities for ...startups, there is not much empirical evidence on the actual hiring behavior of newly founded firms, as observed by Fairlie and